Sensor Correlation Anomaly Detection Without Expert Rules
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing environmental monitoring systems relying on IoT devices struggle with slow and costly implementation of domain knowledge to detect anomalous behavior, leading to underutilization of available data and sub-optimal anomaly detection as the amount of data increases.
Innovation Solution
A computer-implemented method that analyzes correlations between data series from multiple sensors to identify normally correlated data feeds, determining a classification of behavior as normal or anomalous based on these correlations, and uses this knowledge to detect and respond to anomalous events.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If domain knowledge and expert analysis are used to identify relationships between sensor data feeds, then anomaly detection accuracy is improved, but implementation time and cost increase significantly
Solution Approach 1:
The system automatically discovers relationships between data feeds using correlation analysis without requiring expert intervention. The anomaly detection model is trained autonomously on historical data, allowing the system to self-configure and adapt to the specific environment being monitored, thereby eliminating the need for manual domain knowledge incorporation while maintaining high detection accuracy
Solution Approach 2:
The patent replaces the manual mechanical process of expert analysis and rule creation with an automated computational system. Machine learning algorithms automatically identify patterns and relationships in sensor data, substituting human expert work with computational processes that are both faster and more scalable
2Reliability
If domain knowledge and expert analysis are used to identify relationships between sensor data feeds, then anomaly detection accuracy is improved, but implementation cost increases significantly
Solution Approach 1:
The system automatically discovers relationships between data feeds using correlation analysis without requiring expert intervention. The anomaly detection model is trained autonomously on historical data, allowing the system to self-configure and adapt to the specific environment being monitored, thereby eliminating the need for manual domain knowledge incorporation while maintaining high detection accuracy
Solution Approach 2:
The patent replaces the manual mechanical process of expert analysis and rule creation with an automated computational system. Machine learning algorithms automatically identify patterns and relationships in sensor data, substituting human expert work with computational processes that are both faster and more scalable
3Loss of information
If the amount of data from multiple sensors increases, then more relationships and insights become available, but data underutilization occurs due to complexity
Solution Approach 1:
The correlation analysis framework is universally applicable to any combination of sensor data feeds, automatically adapting to the specific sensors and environment. The system handles multiple data sources through a unified approach that identifies relationships without requiring separate analysis for each sensor combination, thereby preventing data underutilization while managing complexity
Solution Approach 2:
The patent replaces the manual mechanical process of expert analysis and rule creation with an automated computational system. Machine learning algorithms automatically identify patterns and relationships in sensor data, substituting human expert work with computational processes that are both faster and more scalable
4Device complexity
If traditional monitoring approaches are used, then implementation is simpler, but anomalies are missed and normal behavior is misclassified
Solution Approach 1:
The system automatically discovers relationships between data feeds using correlation analysis without requiring expert intervention. The anomaly detection model is trained autonomously on historical data, allowing the system to self-configure and adapt to the specific environment being monitored, thereby eliminating the need for manual domain knowledge incorporation while maintaining high detection accuracy
Solution Approach 2:
The patent replaces the manual mechanical process of expert analysis and rule creation with an automated computational system. Machine learning algorithms automatically identify patterns and relationships in sensor data, substituting human expert work with computational processes that are both faster and more scalable
Data Source
AI summary
A computer implemented method of detecting anomalous behaviour within an environment is provided. The environment is monitored by a plurality of sensors providing a plurality of data feeds. Each data feed provides a respective data series representing a respective physical property of the environment over time. The method of detects an occurrence of an event within the environment and identifies a type of that event. The method identifies a plurality of normally correlated data feeds from the plurality of data feeds for the type of the event. The method determines a respective degree of correlation between the respective data series provided by each of the normally correlated data feeds for the occurrence of the event. The method determines a classification of the behaviour within the environment based on the determined degree(s) of correlation, the classification indicating whether the behaviour is normal or anomalous for the environment. Also provided is a method of training an anomaly detector for detecting anomalous behaviour within such an environment, as well as associated computer systems, computer programs, computer-readable data carriers and data carrier signals for performing such methods.